A concerted action of HNF4α and HNF1α links hepatitis B virus replication to hepatocyte differentiation

Authors


*E-mail protzer@tum.de; Tel. (+49) 89 41406886; Fax (+49) 89 41406823.

Summary

Hepatitis B virus (HBV) is an important human pathogen, which targets the liver extremely efficient, gaining access to hepatocytes by a so far unknown receptor and replicating in a hepatocyte-specific fashion. Cell differentiation seems to determine HBV replication. We here show that the level of hepatocyte differentiation, as indicated by hepatocyte polarization and metabolic activity, is closely correlated to the transcription of the HBV RNA pregenome. Pregenome transcription determined the level of HBV replication in various cell lines of hepatocellular origin and in primary human hepatocytes. A variety of hepatocyte-enriched nuclear factors have been described to regulate transcription of the pregenome, but it remained unknown which factors link HBV replication to hepatocyte differentiation. We determined that high expression levels of HNF4α but not its potential cofactors or other hepatocyte-enriched transcription factors were essential for efficient HBV replication, and link it to hepatocyte differentiation. HNF1α contributed to the control of HBV replication because it regulated the expression of HNF4α. Thus, a concerted action of HNF4α and HNF1α, which also determines morphological and functional differentiation of hepatocytes, links HBV replication to hepatocyte differentiation.

Introduction

Human hepatitis B virus (HBV) transiently and persistently infects the liver, which may result in an inflammatory liver disease – acute or chronic hepatitis B. Chronic hepatitis B often results in liver cirrhosis or hepatocellular carcinoma (HCC). HBV is an enveloped virus whose small (3.2 kb), relaxed circular (rc) DNA genome shows an extremely compact organization. The genomic viral DNA is encapsidated together with the viral polymerase within an icosahedral nucleocapsid consisting of the HBV core protein. The viral envelope is densely packed with large (L), middle (M) and – predominantly – small (S) viral envelope proteins. In addition, the virus encodes the regulatory protein X (for review see Seeger and Mason, 2000; Ganem and Schneider, 2001). Like all hepadnaviruses, HBV multiplies its genomes by reverse transcription of an RNA pregenome (Summers and Mason, 1982).

Following infection, the rcDNA genome is imported into the hepatocyte nucleus where it is completed to a covalently closed cirular DNA (cccDNA). cccDNA serves as the viral transcription template containing four unidirectional, overlapping open reading frames. Four promoters in concert with enhancers I and II (Shaul et al., 1985; Yee, 1989; Su and Yee, 1992) control transcription of independent mRNAs which all remain unspliced: preC/C, preS1, preS2/S and X (summarized in Ganem and Schneider, 2001). The 0.9 kb X (X protein), 2.1 kb preS2/S (M and S protein) and 2.4 kb preS1 (L protein) are subgenomic RNAs and serve as mRNAs. preC/C RNA (3.5 kb) consists of a subset of more than genome-length RNAs, which are transcribed from two physically overlapping but functionally distinct promoters (Yaginuma and Koike, 1989; Yu and Mertz, 1996). C RNA can serve as mRNA for translation of HBV core and polymerase proteins or as pregenomic RNA (pgRNA), which is encapsidated and reverse transcribed in the viral capsid (Ganem and Schneider, 2001). preC mRNA is only translated into the hepatitis B e antigen (HBeAg) (Nassal et al., 1990).

Hepatitis B virus transcription by host RNA-polymerase II is regulated finely by hepatocyte-enriched transcription factors: CCAAT/enhancer-binding protein (C/EBP) (Lopez-Cabrera et al., 1990; 1991) and hepatocyte nuclear factors (HNF), such as HNF1 (Raney et al., 1990), HNF3 (Ori and Shaul, 1995) or HNF4 (Raney et al., 1997; Tang and McLachlan, 2001) in concert with nuclear receptors PGC-1α (Shlomai et al., 2006) and COUP-TF (Raney et al., 1997; Yu and Mertz, 2003). These factors confer hepatocyte-specific activity of the viral preS1 (Lopez-Cabrera et al., 1991; Raney and McLachlan, 1995; Raney et al., 1995) and preC/C promoters (Lopez-Cabrera et al., 1990; Johnson et al., 1995; Raney et al., 1997; Tang and McLachlan, 2001; 2002; Zheng et al., 2004; Shlomai et al., 2006), and viral enhancer elements (Yee, 1989; Yuh and Ting, 1993), and determine the hepatocyte tropism of HBV in addition to a yet unknown receptor on the cell surface.

Hepatocyte-enriched transcription factors also play an important role in liver development and hepatocyte differentiation. It has recently been shown that HNF4α is essential for morphological and functional differentiation of hepatocytes, whereas expression of HNF1α is not an absolute requirement for mammalian liver development (Li et al., 2000; Hayhurst et al., 2001). HNF4α also is the dominant regulator of the epithelial phenotype of hepatocytes and thus is essential for a normal liver architecture (Parviz et al., 2003). Forty-two per cent of the genes occupied by RNA polymerase II in hepatocytes are bound by HNF4α (Odom et al., 2004). Primary human hepatocytes (PHH), which per se are highly differentiated, and hepatoma cells cultivated under differentiating conditions, have been reported to better support HBV infection and replication (Gripon et al., 1989; 2002; Glebe et al., 2001; Sprinzl et al., 2001). In addition, the activity of preC/C – and especially preS1 – promoters seems to depend on the hepatocyte differentiation state and is enhanced in quiescent hepatocytes (Chang and Ting, 1989; Raney et al., 1990; Ozer et al., 1996).

However, none of these studies analysed in detail which hepatocellular factors link HBV transcription and replication to hepatocyte differentiation. Considering hepatocyte-enriched transcription factors as putative candidates, it is not known whether only one or a combination of these transcriptional regulators is responsible for differentiation-dependent activity of HBV promoters. Moreover, it is still unclear whether hepatocyte-enriched transcription factors act on HBV promoters in a dose-dependent manner. Thus, the essential link between the efficiency of HBV replication and the degree of hepatocyte differentiation remained to be elucidated.

In this study, we provide evidence that (i) HBV replication is highly dependent on hepatocyte differentiation state with transcription of pgRNA being the limiting step, and (ii) among all hepatocyte-enriched transcription factors, HNF4α is the essential link between the efficiency of HBV replication and degree of hepatocyte differentiation, regulating the former in a dose-dependent manner.

Results

Markers of hepatocyte differentiation

To analyse the dependence of HBV replication on hepatocyte differentiation, we chose PHH, hepatoma cell lines HepG2, HuH7 and HepaRG and hepatocyte cell line Pop10. HepaRG cells, which differentiate and become permissive for HBV if cultivated with corticosteroids and dimethyl sulfoxide (DMSO) (Gripon et al., 2002; Engelke et al., 2006), were compared in an undifferentiated and a differentiated state. To prove the differentiation status of our cells, we characterized the expression of hepatocyte-specific differentiation markers.

All cells stained positive for CK8/18 (data not shown), and expression levels of neither albumin nor ferritin (Fig. 1A) correlated with the expected level of hepatocyte differentiation. Therefore, we analysed key proteins of metabolic pathways in hepatocytes. Expression levels of cytochrome p450 family member CYP1A2, steroid regulatory element-binding protein 2 (SREBP-2), apolipoprotein B (ApoB 100) (Fig. 1A), 2′3′-tryptophandioxygenase (TDO) and cytoplasmic phosphoenolpyruvate carboxykinase (PEPCK) (Fig. 1B) were markedly higher in primary hepatocytes than in HepG2 or HuH7 cells and increased upon differentiation in HepaRG cells. All cells expressed comparable levels of hydroxymethylglutaryl-CoA-reductase and pterin-4 alpha-carbinolamine dehydratase (data not shown). In addition, we determined expression of mitochondrial cytochromes (liver-specific antigen, LSA). Only in primary hepatocytes and differentiated HepaRG cells expression of LSA was detected (Fig. 1A).

Figure 1.

HBV replication and hepatocyte differentiation markers in cells of hepatocyte origin.
A. Western blot analysis of OATP- C, CYP1A2, LSA, albumin, ferritin, SREBP-2, ApoB100 and β-actin. Molecular weights of corresponding proteins are given in kDa.
B. Expression levels of TDO, PEPCK, OATP-C and BSEP were determined relative to GAPDH by LightCycler™ real-time reverse transcription polymerase chain reaction (RT-PCR). The normalized expression level in PHH was set to 1. Mean values ± SD from three different samples of each cell line are given (except HepaRG: two samples each). PHH from four different donors were included.
C. Western blot analysis of HBV core protein and GFP.
D. PHH, HepG2, Huh7, Pop10 and differentiated and undifferentiated HepaRG cells were transduced with Ad-HBV to induce HBV replication. Enveloped HBV progeny released from transduced cells was quantified by DNA dot blot analysis following CsCl gradient sedimentation. Numbers of enveloped HBV particles released per cell were calculated. In (A), (C) and (D) representative experiments are shown.
E. HBV pgRNA at day 7 after transduction with Ad-HBV was determined relative to GAPDH by LightCycler™ real-time RT-PCR. The normalized expression level in PHH was set to 1. Mean values ± SD from two independent experiments are given.

As polarization is an important feature of hepatocytes and other epithelial cells, we exemplarily analysed organic anion transporter polypeptide C (OATP-C) and bile salt exporting pump (BSEP) localized in basolateral and apical hepatocyte plasma membranes respectively. Western blotting detected OATP-C only in primary hepatocytes (Fig. 1A), in which basolateral distribution and thus polarization was confirmed by immunofluorescence staining (data not shown). Levels of OATP-C and BSEP mRNA were the highest in PHH, and increased upon differentiation in HepaRG cells indicating regulation mainly at the level of transcription (Fig. 1B). In Pop10 cells, no TDO mRNA and only minimal amounts of PEPCK, OATP-C and BSEP mRNA were detected.

Taken together, differentiated hepatocytes or hepatoma cells expressed a set of proteins indicating high metabolic activity as well as hepatocyte polarization. Expression of TDO, PEPCK, BSEP and OATP-C proved most appropriate as indicators of hepatocyte differentiation as their expression levels (i) were markedly higher in PHH than in hepatoma cells and strongly increased upon differentiation of HepaRG cells, and (ii) were undetectable in non-hepatocytes and very low or undetectable in hepatic cells not replicating HBV (Pop10). Notably, all differentiation markers were regulated primarily at the transcriptional level.

Efficiency of HBV replication in different hepatocytes

To study the efficiency of HBV replication independent of the early steps of virus uptake, we transduced PHH, HepG2, HuH7, differentiated and undifferentiated HepaRG and Pop10 cells with an HBV genome and a GFP cDNA using adenoviral vector Ad-HBV. Comparable (90–95%) transduction rates were confirmed by fluorescence microscopy and Western blot analysis of GFP expression (Fig. 1C).

As shown in Fig. 1D, PHH started to release enveloped progeny HBV particles at day 3, HepG2 and HepaRG cells at day 4 and HuH7 cells at day 5 after transduction with Ad-HBV. Pop10 cells did not release progeny HBV at all (detection limit: one HBV particle per cell). PHH secreted with > 500 virions per cell per day the highest amounts, markedly more than HepG2 cells (13.6-, 13.4-, 2.7-fold, at day 5, 6 and 7 post transduction respectively), differentiated HepaRG cells (14.4-, 14.3-, 3.9-fold) or HuH7 cells (10.5-, 15.7-, 13-fold). Differentiated HepaRG cells released 2- to 3.3-fold more HBV progeny than respective undifferentiated cells.

By analysing the HBV replication cycle, we found expression levels of 3.5 kb HBV RNAs containing the HBV pregenome (pgRNA) to be the limiting step. Although transduced to equal levels, HepG2, differentiated HepaRG and HuH7 cells expressed only 9%, 8.5% and 6% of the amount of HBV pgRNA detected in PHH at day 7, respectively, and Pop10 cells no HBV pgRNA at all (Fig. 1E). Upon differentiation, HBV pgRNA increased 5.7-fold in HepaRG cells. Accordingly, expression of HBV core protein was the highest in primary hepatocytes and increased upon differentiation of HepaRG cells (Fig. 1C).

Taken together, primary hepatocytes, the highest differentiated cells in our study, replicated HBV better than hepatoma cells and released more progeny HBV. Pop10 cells, which express none or little of studied hepatocyte differentiation markers, did not replicate HBV at all. The differences in the efficiency of HBV replication depended largely on the expression level of the HBV pregenome. Thus, efficiency of HBV replication was regulated at the transcriptional level.

Cells efficiently replicating HBV express high levels of hepatocyte nuclear transcription factors HNF1α, HNF3γ and HNF4α

As HBV replication as well as hepatocyte differentiation was primarily regulated at the transcriptional level, we analysed hepatocyte-specific transcription factors HNF1α/β, HNF3α/β/γ, HNF4α, C/EBPα/β, LRH-1, PGC-1α and COUP-TF II as candidate common denominators.

Primary human hepatocytes expressed the highest levels of HNF1α, HNF3γ and HNF4α. Upon differentiation, HepaRG cells expressed increasing amounts of HNF1α, HNF3γ, HNF4α and C/EBPα and β, but decreasing amounts of HNF3α. In addition, PHH as well as HepaRG cells expressed different isoforms of C/EBPβ than the other cells analysed. Pop10 cells, in contrast, only contained trace amounts of HNF3γ and HNF4α and no detectable HNF1α (Fig. 2A and B).

Figure 2.

Expression patterns of liver enriched transcription factors.
A. Western blot analysis of HNF1α, 3α, 3β, 3γ and 4α, C/EBPα and β, LRH-1, PGC-1α and COUP-TF II expression in primary human hepatocytes (PHH), HepG2, Huh7, Pop10, differentiated and undifferentiated HepaRG cells. Molecular weights of corresponding proteins are given in kDa.
B. LightCycler™ real-time RT-PCR analysis of HNF1α, 3γ and 4α expression. Normalized expression levels in PHH were set to 1. Mean values ± SD from three different samples of each cell line are given (except HepaRG: two samples each). PHH from four different donors were included.

To correlate transcription factor expression levels with HBV replication, we cultivated HBV-producing cell lines under conditions promoting differentiation: a collagen matrix and medium containing dexamethasone, DMSO and low FCS. Levels of OATP-C and BSEP mRNA rose 2.9- and 5.7-fold in HepG2.2.15 and 11.2- and 6.4-fold in HepG2-H1.3 cells, respectively, indicating successful differentiation during long-term culture (data not shown). In parallel, HNF4α, HNF1α and HNF3γ levels rose 3.2-, 3.3- and 4.6-fold in HepG2.2.15 cells and 7.0-, 4.5- and 2.6-fold in HepG2-H1.3 cells (Fig. 3C), respectively, whereas C/EBPα/β, HNF3α, HNF3β, LRH-1, PGC-1α and COUP-TF II expression did not show a clear tendency (data not shown). To exclude an influence of the cell culture conditions on the overall synthetic capacity of our hepatoma cells, we normalized expression levels of HNF4a to those of albumin, the major synthetic product of hepatocytes (Fig. 3D).

Figure 3.

HBV replication and hepatocyte differentiation in stably HBV-producing cell lines. Stably HBV-producing cell lines HepG2 2.15 (left) and HepG2-H1.3 (right) were cultured under differentiating conditions and lysed at indicated time points. A representative, parallel experiment is shown. Values obtained at day 0 were set to 1.
A. HBV-DNA was quantified by LightCycler™ real-time PCR in 200 μl of each of three parallel cell culture medium samples relative to an external standard. The amount of HBV progeny per cell was calculated (given as mean values ± SD) (grey bars). HBV pgRNA was determined in three parallel samples relative to GAPDH by LightCycler™ real-time RT-PCR (black line).
B. Western blot analysis of HBV core and L protein as well as albumin and β-actin. Molecular weights are given in kDa.
C. HNF1α, HNF3γ and HNF4α content was quantified relative to β-actin by chemiluminescence imaging of Western blots.
D. HNF4α mRNA was quantified by LightCycler™ real-time RT-PCR and is given in comparison with the protein quantified relative to albumin by chemiluminescence imaging of Western blots.

Accordingly, levels of HBV pgRNA increased 8.1-fold in HepG2.2.15 and 8.6-fold in HepG2-H1.3 cells. The amount of progeny HBV released into the cell culture medium increased from day 0 to 16 from 2.3 ± 0.14 to 217.4 ± 43.38 (HepG2.2.15 cells) and from 1.79 ± 0.1 to 1281.8 ± 149.4 HBV-DNA copies per cell (HepG2-H1.3 cells) (Fig. 3A). HBV core and L protein also increased over time: 10- and 7.1-fold in HepG2.2.15 cells and 14- and 8.8-fold in HepG2-H1.3 cells respectively (Fig. 3B).

Taken together, differentiation of the two HepG2-based cell lines resulted in increasing expression of HNF1α, HNF3γ and HNF4α and rising HBV replication. In both cell lines, HBV replication depended on transcription of HBV pgRNA which correlated most closely with rising levels of HNF4α. This led us to the hypothesis that HNF1α, HNF3γ and/or HNF4α provide the essential link between hepatocyte differentiation and HBV replication.

High expression levels of HNF4α and HNF1α are required for efficient HBV replication

To test whether efficient HBV replication depends on high expression levels of HNF4α, HNF1α or HNF3γ, we performed knock-down experiments in HepG2-H1.3 cells using specific siRNAs for these transcription factors. As control, we used HepG2-H1.3 cells transfected with AlexaFluor488-labelled, non-silencing siRNA to which all effects of specific siRNAs were related.

We achieved a long-lasting 50%, 67% and 82% knock-down of HNF1α, HNF3γ and HNF4α respectively (Fig. 4A). At day 5 after the knock-down of HNF4α and HNF1α, mRNA levels of OATP-C decreased 2.55 ± 0.23- and 2.30 ± 0.13-fold, of BSEP 1.50 ± 0.16- and 1.30 ± 0.07-fold, of TDO 2.60 ± 0.01- and 1.75 ±  0.19-fold, and of PEPCK 2.80 ± 0.19 and 2.30 ±  0.13-fold (Fig. 4B).

Figure 4.

Knock-down of HNF4α, HNF1α and HNF3γ and its effect on HBV replication and expression of hepatocyte-specific differentiation markers. HepG2-H1.3 cells were transfected with 5 nM siRNA (+) specific to either HNF4α, HNF1α or HNF3γ or non-silencing (ns). Cells were harvested on day (d) 3, 5 and 7 after transfection as indicated.
A. Western blot analysis of HNF4α, HNF1α and HNF3γ. Knock-down efficiency of specific siRNAs is given in percentage (relative to ns siRNA).
B. Normalized expression levels of OATP- C, BSEP, TDO and PEPCK were determined by LightCycler™ real-time RT-PCR relative to delta aminolevulinate synthase (ALAS).
C. Normalized expression levels of HBV pgRNA were determined by LightCycler™ real-time RT-PCR relative to delta aminolevulinate synthase (ALAS).
D and E. (D) Western blot analysis of HBV core protein at days 5 and 7 and (E) HBV L protein at day 5.
F. HBV-DNA was quantified by LightCycler™ real-time PCR relative to an external standard. The amount of HBV progeny per cell was calculated.
G. Southern blot analysis of HindIII-digested total cellular DNA using a 32P-labelled HBV-DNA probe. HBV replicative intermediates were normalized to HBV integrates following phosphoimager quantification.
H. PCR amplification products of HBV cccDNA and mitochondrial DNA. Ratios were determined from real-time PCR analysis. Parental HepG2 cells were used as control.
In (B) and (C) ns siRNA-transfected cells were set to 1. In (A), (D), (E), (G) and (H) representative experiments are shown. In (B), mean values ± SD of two, in (C) and (F), mean values ± SD of three independent experiments are given; *P < 0.05, **P < 0.01, Student's t-test.

Transcription of HBV pgRNA was significantly reduced in three independent experiments after knock-down of HNF4α (4.7 ± 0.3-fold at day 3; 2.2 ± 0.1-fold at day 5 post transfection) and HNF1α (4.05 ± 0.25-fold at day 5) (Fig. 4C). HBV core protein was diminished after the knock-down of HNF4α, HBV L protein after the knock-down of HNF1α (Fig. 4D and E), both leading to a significant decrease in HBV progeny release (Fig. 4F). Unlike HNF1α and HNF4α, HNF3γ knock-down did not have any inhibitory effect (Fig. 4C–F).

Southern blot analysis showed a 70% and 75% diminished HBV replication after the knock-down of HNF1α and HNF4α respectively (Fig. 4G). Concomitantly, the accumulation of HBV cccDNA in the nucleus of HepG2-H1.3 cells was inhibited (inhibition by HNF1α 88%, HNF4α 76%, Fig. 4H). As these siRNAs induced neither interferon-γ-inducible protein-10 (IP-10) nor 2′-5′-oligoadenylate-synthetase, we excluded that induction of interferon type I as a possible side-effect of siRNAs affected HBV replication (data not shown).

To show the direct effect of HNF4α on HBV replication in a given cell, we performed immunostaining for HNF4α and HBV core protein at day 7 after transfection with siRNA for HNF4α or with non-silencing siRNA (Fig. 5). In cells with reduced or undetectable HNF4α, HBV core protein was also strongly reduced. From these experiments, we concluded that high expression levels of HNF4α and HNF1α are needed for efficient HBV replication as well as expression of hepatocyte differentiation markers.

Figure 5.

Analysis of HBV core and HNF4α expression in single cells. Immunofluorescence staining of HNF4α (AlexaFluor488™, green), HBV core protein (AlexaFluor568™, red) or cell nuclei (DAPI, blue) in cells transfected with either HNF4α-specific or ns siRNA at day 7 post transfection. Laser scanning confocal microscopy, scale bar 60 μm.

Levels of HNF4α determine HBV core protein expression and HBV replication in vivo in human tumour and non-tumour liver tissue

To test whether expression levels of hepatocyte-enriched transcription factors also determine the efficiency of HBV replication in vivo in livers of infected patients, we analysed expression of hepatocyte-enriched transcription factors, production of HBV core protein and pgRNA in tumour as well as corresponding peritumour tissues of HCC patients chronically infected with HBV. Tumours were graded as shown in Table 3. Minor differences of either hepatocyte-enriched transcription factors or HBV core protein (Fig. 6A) or HBV pgRNA (Fig. 6B) were detected between tumour and peritumour tissue samples of a given patient, whereas we found a high interindividual variation. In all samples analysed, amounts of HNF4α significantly correlated with amounts of HBV core protein (Pearson correlation 0.82, P < 0.01) (Fig. 6A). Also, levels of pgRNA correlated with expression levels of HNF4α (Pearson correlation 0.57, P = 0.057) (Fig. 6B). Thus, the data confirm in vivo in patient material that efficient HBV replication relies on high expression levels of HNF4α.

Table 3.  Staging and grading of hepatocellular carcinoma samples.
Patient numberSerum HBsAg/anti-HBcTumour staging and gradingPeritumour, fibrosis stagePeritumour, inflammation grade
  1. nd, not determined.

1+pT3 N0 Mx G2 R033
2+pT1 N0 Mx G2-4 R02–32
areals of mixed (high and no) differentiation
3+pT1 N0 Mx G2 R02–32
4+pT1 N0 Mx G2 R0ndnd
5+pT2 N0 Mx G2 R02–32
6+pT1 N0 Mx G1 R042
7+pT3 N0 Mx G3 R041
8+pT1 N0 Mx G1 R041
Figure 6.

Correlation of the amounts of HNF4α and HBV core protein in tumour (T)–peritumour (P) tissues of HCC patients chronically infected with HBV.
A. Western blot analysis for HNF4α and HBV core protein in tumour–peritumour tissue samples from five patients (pt 1–5), three unpaired HCC tissue samples (pt 6, 7, 8) and three normal liver tissue samples (pooled). HNF4α and HBV core protein, respectively, were quantified relative to albumin using chemiluminescence imaging. Relative band intensities are given.
B. Normalized expression levels of HBV pgRNA were determined by LightCycler™ real-time RT-PCR relative to GAPDH and are shown in comparison with HNF4α expression.

Discussion

In this study, we demonstrated that (i) HBV replication is highly dependent on the degree of hepatocyte differentiation, that (ii) pgRNA transcription determines the efficiency of HBV replication in relation to hepatocyte differentiation in cultivated cells as well as in liver tissue, and that (iii) HNF4α is the key regulator of HBV replication in context of hepatocyte differentiation. Not the presence of HNF4α alone, but its intracellular amounts determined the efficiency of HBV pregenome transcription and thus HBV replication in hepatoma cell lines as well as in primary human hepatocytes. This was confirmed when liver tissue samples from HBV-infected individuals were studied. As HNF4α also regulated morphological and functional differentiation of hepatocytes, it linked efficient HBV replication to hepatocyte differentiation.

We identified transcription of the HBV pgRNA to be the crucial step in the viral life cycle. Among all hepatocyte-enriched transcription factors studied, only HNF4α proved essential in controlling intracellular HBV replication. HNF4α regulated transcription of the HBV pregenome as a function of the hepatocellular differentiation state. Tang and McLachlan (2001) had reported earlier that a concerted action of nuclear hormone receptors including HNF4α is needed to induce transcription of HBV pgRNA and replication of HBV in non-hepatocytes. In cells of hepatocyte origin, where hepatocyte-enriched transcription factors are constitutively expressed, we found constant levels of nuclear hormone receptor PGC1α apparently sufficient to serve as a cofactor for HNF4α. In contrast, we found a strong positive correlation between intracellular amounts of HNF4α and efficiency of viral replication. COUP-TF competing with HNF4α for the same binding site (Raney et al., 1997; Yu and Mertz, 2003) was present in all cells at constant levels. This explained why a threshold of HNF4α expression had to be reached, above which HBV replication depended on the level of HNF4α.

Hepatitis B virus replication rises when HBV-producing cells are kept under differentiating conditions (Ozer et al., 1996): plating onto extracellular matrix, culture medium containing dexamethasone and DMSO but low FCS (Glebe et al., 2001 and Fig. 3). Dexamethasone enhances HNF4 and HNF1 expression via an upstream steroid responsive element (Bailly et al., 2001). DMSO induces cellular differentiation by a so far unknown mechanism, and enhances HBV replication (Gripon et al., 1989; 2002; Glebe et al., 2001). Furthermore, it regulates histone acetylation and methylation (Sarg et al., 2005), which may render target sites in the viral genome accessible to hepatocyte nuclear factors.

It was only recently that HNF4α was pinpointed to be the key regulator of morphological and functional differentiation of hepatocytes, essential for metabolic function and formation of a polarized hepatic epithelium (Parviz et al., 2003) as well as for the formation of cell–cell contacts (Battle et al., 2006). HNF1α primarily regulates hepatocyte polarization (Sakaguchi et al., 2002). Accordingly, we found that HNF4α and HNF1α knock-down also affected expression of bile acid transporters OATP-C and BSEP, which are only expressed in highly differentiated, polarized hepatocytes, and metabolic enzymes such as SREBP-2, PEPCK and TDO, which indicated high metabolic activity.

Although HNF4α was the key player in regulating transcription of the HBV pregenome in our experiments, knock-down of HNF1α also influenced HBV replication and progeny virus release. As HNF1α is essential for expression of viral envelope protein L (Courtois et al., 1988), it controls release of viral particles from infected cells. Its influence on viral pgRNA transcription and viral replication, however, was unexpected as neither the overlapping HBV pre-core/core promoter/enhancer II nor the upstream enhancer I, which control pgRNA transcription, contain HNF1α binding sites unless they are mutated (Gunther et al., 1996).

As HNF1α is essentially involved in the control of HNF4α expression (Bailly et al., 2001; Odom et al., 2004), we speculated that its knock-down diminished HBV replication by affecting the transcription of HNF4α. Indeed, we observed a substantial decrease in the amounts of HNF4α at days 3 and 5 after knock-down of HNF1α (data not shown), coincident with downregulation of HBV pgRNA transcription during HNF1α knock-down.

However, in HBV transgenic mice, transcription of the HBV pregenome was not measurably affected by a knock-out of HNF1α, and intracellular HBV replication was even slightly increased (Raney et al., 2001). As the effect of HNF1α on expression of HNF4α as well as on HBV replication ceased after 5 days in our experiments, we suppose that hepatocytes compensate for the lack of HNF1α.

Guidotti et al. (1997) reported that HBV replication per cell remained constant after partial hepatectomy in livers of HBV transgenic mice. This does not argue against a strong dependence of HBV replication on a highly differentiated hepatocyte, because liver regeneration after partial hepatectomy takes place by proliferation of highly differentiated hepatocytes without shortening of the G1-phase (Fausto and Campbell, 2003) or reduction of HNF4α expression levels (Flodby et al., 1993).

Although we clearly demonstrated that HNF4α and to a lesser extent HNF1α link HBV replication to hepatocyte differentiation by controlling transcription of its RNA pregenome, we cannot exclude an additional influence of hepatocyte metabolic functions controlled by these transcription factors (Naiki et al., 2002; Parviz et al., 2003; Odom et al., 2004).

Taken together, we have shown that HBV replication strongly depends on the degree of hepatocyte differentiation, and that high HNF4α expression levels required for development of a highly differentiated hepatocyte link efficient HBV replication to hepatocyte differentiation.

Our results provide new insights into virus–host cell interaction that will be helpful for the generation of new models of HBV infection and for the development of therapeutics against HBV. Furthermore, the results of this study should be taken into consideration when prescribing drugs, e.g. phenobarbital (Bell and Michalopoulos, 2006) increasing nuclear expression of HNF4α to patients with hepatitis B.

Experimental procedures

Cell culture

Primary human hepatocytes were isolated from surgical liver resections after receiving informed consent of the donor. After a two-step collagenase perfusion of tissue remnants and subsequent differential centrifugation, PHH were plated onto collagen-coated dishes and cultivated as described (Schulze-Bergkamen et al., 2003). Kidney cell line HEK293, HCC cell lines HepaRG, HepG2 and HuH7 as well as hepatocyte cell line Pop10 were cultivated as described (Sprinzl et al., 2001; Gripon et al., 2002; Nguyen et al., 2005). HepG2.2.15 cells, an established cell line replicating HBV from four intergrated dimeric HBV genomes (Sells et al., 1987), and HepG2-H1.3 cells, a cell line containing one copy of a 1.3-fold overlength HBV genome, which establishes HBV cccDNA as additional transcription template (Jost et al., 2007; Protzer et al., 2007), were plated onto collagen-coated dishes, and after achieving 80–90% confluence maintained in PHH medium/Dulbecco's modified Eagle's medium (1:1), containing 1% bovine serum.

Induction and analysis of HBV replication

To induce HBV replication, AdHBV, a first generation adenoviral vector (Ad5ΔE1/E3) containing a 1.3-fold HBV genome and a GFP expression cassette, was used (Sprinzl et al., 2001). Cells were transduced with AdHBV to achieve 90–95% green fluorescent cells. For PHH, we used a multiplicity of infection of 5, for HuH7 and HepaRG of 30 and for HepG2 and Pop10 cells of 10 infectious units per cell.

Hepatitis B virus particles contained in 2.5 ml of cell culture medium were sedimented through a caesium chloride step gradient (density 1.15–1.4 g ml−1). Enveloped, DNA-containing HBV particles were identified by dot blot hybridization of respective fractions using a 32P-labelled HBV-DNA probe (Jost et al., 2007; Protzer et al., 2007), and quantified using a PhosphoImager (Bio-Rad Laboratories, Munich, Germany) relative to a dilution series of HBV-DNA.

Total cellular DNA (15 μg) was digested with HindIII, which excises the HBV integrate, separated through a 1.5% agarose gel, and analysed by Southern Blotting using a 32P-labelled HBV-DNA probe as described (Sprinzl et al., 2001). HBV replicative intermediates were quantified relative to the HBV integrate.

Real-time PCR

Total DNA was extracted from cell culture medium using QIAamp MiniElute Virus Spin Kit and from cells using DNeasy kit (Qiagen), and HBV-DNA including cccDNA was quantified as described (Untergasser et al., 2006).

Total RNA was extracted using RNeasy® total RNA extraction kit (Qiagen, Hilden, Germany). One microgram of total RNA was transcribed into cDNA after DNase digestion using SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, USA) for PCR. HBV pgRNA was detected as described previously (Untergasser et al., 2006). For gene expression analysis, appropriate exon–exon spanning primer pairs were selected whenever possible. Primer positions are listed in Table 1. Real-time PCRs were performed using the LightCycler™ system and normalized to a dilution series of calibrator cDNA using the Relative Quantification Software (both Roche Diagnostics, Mannheim, Germany) as described in detail (Untergasser et al., 2006).

Table 1.  Primers used for LightCycler™ real-time PCR.
Target geneGenBank accession No.Primer forwardPrimer reverse
Organic anion transporter polypeptide C (OATP-C)NM_0064461939–19542038–2021
Bile salt-exporting pump (BSEP)NM_0037423577–35923901–3886
2′3′-Tryptophan dioxygenase (TDO)NM_005651921–9361130–1114
Pterin-4 alpha-carbinolamine dehydratase (PDG)NM_000281316–332519–503
Glyceraldehyde-3-phosphate-dehydrogenase (GAPDH)NM_002046607–623973–958
Delta-aminolevulinic acid synthetase (ALAS)NM_0006882007–20262192–2177
Cytosolic phosphoenolpyruvate carboxykinase (PEPCK)NM_0025911626–16431840–1824
Hepatocyte nuclear factor (HNF)4αNM_000457687–702962–945
HNF1αNM_000545980–9951309–1292
HNF3γNM_004497327–344571–550
Interferon-gamma inducible protein −10 (IP-10)NM_001565145–160310–294
2′5′-Oligoadenylate synthetase (2′5′ OAS)NM_016816377–392601–585

Protein expression analysis

For extraction of total cellular proteins CHAPS buffer (10 mM Hepes, pH 7.4, 150 mM NaCl, 1% Chaps), for extraction of proteins from tissue samples SDS buffer (15 mM Tris HCl, pH 6.8, 2.5% glycerol, 0.5% SDS, 1 mM EDTA) after addition of protease inhibitors was used. For detection of transporters, membrane proteins were extracted with TED buffer (50 mM Tris, 1 mM DTT, 1 mM EDTA). For Western blot analysis, cellular lysates containing 15 μg of tissue lysates containing 30 μg of protein per lane were separated by SDS-PAGE (6.5–12.5% according to the expected protein size), transferred onto nitrocellulose membranes, stained with appropriate primary (see Table 2) and secondary antibodies (Sigma, Munich, Germany) and detected and quantified by enhanced chemiluminescence (WestDura, Pierce, Rockford, USA) using the Gel Doc 2000 System (Bio-Rad Laboratories, Munich, Germany). For reprobing, membranes were treated with 0.2 N NaOH for 10 min at room temperature before staining with the new antibody.

Table 2.  Primary antibodies used for Western blot analysis.
TargetSourceWorking dilutionManufacturer
Liver-specific antigen (LSA)Mouse1:1000Clone OCH1E5, DAKO
Organic anion transporter polypeptide C (OATP-C)Mouse1:1000Clone mMDQ (provided by D. Keppler)
β-ActinMouse1:4000Clone AC-15, Sigma
Hydroxyl-methylglutaryl-CoA-reductase (HMG-CR)Goat1:400Polyclonal antiserum, Santa Cruz
Steroid regulatory element binding protein – 2 (SREBP-2)Mouse1:5000Clone IgG-1C6, BD Biosciences
Cytochrome p450 family member (CYP1A2)Rabbit1:1000Polyclonal antiserum, BD Biosciences
Apolipoprotein B (ApoB)Mouse1:1000Clone 13, BD Biosciences
AlbuminRabbit1:2000Polyclonal antiserum, DAKO
FerritinRabbit1:1000Polyclonal antiserum, Santa Cruz
Hepatocyte nuclear factor (HNF)4αRabbitAll 1:400Polyclonal antisera, Santa Cruz
 HNF1Rabbit
 HNF3αRabbit
 HNF3βRabbit
 HNF3γGoat
 CCAAT/enhancer binding protein (C/EBP)αRabbit
 C/EBPβRabbit
 PGC-1αRabbit
 ARP-1Goat
Liver receptor homologue – 1 (LRH-1)Mouse1:400Clone H2325, R&D
HBV core and GFPRabbit1:2000Polyclonal antiserum (Kratz et al., 1999)
HBV core protein (H800)Rabbit1:10000Polyclonal antiserum (Nassal et al., 1990)
HBV L and M (H863)Rabbit1:1000Polyclonal antiserum (Engelke et al., 2006)

For immunostaining, cells were fixed with 3.7% formaldehyde and stained using monoclonal mouse anti-human HNF4α (Abcam, Cambridge, UK) and polyclonal rabbit anti-HBV core antibodies. Nuclei were stained with Diammino-2-phenylindol (DAPI). Fluorescence images were acquired using confocal microscope FluoView1000 (Olympus, Hamburg, Germany).

Knock-down using siRNAs

Synthetic siRNA against HNF4α (aacctagagattgttacagaa), HNF1α (caggacaagcatggtcccaca), HNF3γ (ttgatggatgttattggctaa) or non-silencing control (aattctccgaacgtgtcacgt) labelled with AlexaFluor488™ and transfection reagent HiPerFect was purchased from Qiagen, Hilden, Germany. A total of 3.5 × 105 HepG2-H1.3 cells per well were seeded onto collagen-coated six-well plates and transfected with siRNAs (5 nM) using the fast-forward protocol provided by the manufacturer as described (Protzer et al., 2007). Transfection efficiency was controlled by fluorescent microscopy of AlexaFlour488™-labelled siRNA. Knock-down efficiency was determined by quantitative Western blot analysis (see above).

Patient samples

Human HCC or surrounding, non-tumorous (peritumour) HBV-infected liver tissue samples were selected from the tissue bank of the Institute of Pathology, University Hospital Cologne, established after informed consent of patients. Selection criteria were: active HBV infection (HBsAg, anti-HBc and/or HBV-DNA positive in patient's serum), absence of any other obvious cause for HCC (e.g. HCV infection, haematochromatosis) and availability of snap frozen tumour and peritumour tissue. Tumours were graded according to the American Joint Commission on Cancer. For detailed information see Table 3. Healthy liver tissue was obtained from human liver grafts (HBV, HCV, HIV negative) not suited for transplantation.

Statistical analysis

The results were analysed using Student's t-test, anova and Pearson correlation. All data are expressed as a mean ± standard deviation. A P-value of 0.05 or less was considered significant.

Acknowledgements

The authors thank Stephan Urban for help with cultivation of HepaRG cells and providing antiserum H863 against HBV preS, Andreas Untergasser for introduction into primary hepatocyte cultures and adenovirus techniques, Dieter Keppler for anti-human OATP-C antibody, Gisela Holz and Heike Oberwinkler for excellent technical assistance, and Martin Kroenke for his continuous support. We thank Didier Trono and Tuan Huy Nguyen for providing Pop10 cells, and Christian Trepo and Christine Guguen-Guillouzo for providing HepaRG cells. The work was supported by the Medical Faculty of the University of Cologne (to U.P.), by grants from Köln Fortune (to M.Q.), the Deutsche Forschungsgemeinschaft Grant PR618/4 and SFB 670, TP9 (to U.P.), and the Doerenkamp Stiftung (to T.G.). The tumour bank was supported by the German Ministry of Education and Research (German Competence Network for Viral Hepatitis, Grant 01KI0405).

Ancillary